Gary L. Wells

Using staged-crime experiments for unsuspecting people, Wells' experiments have shaped a scientific understanding of problems with the reliability of eyewitness identification evidence and the role that poor lineup procedures play in contributing to mistaken eyewitness identification and false confidence on the part of eyewitnesses.

Wells introduced the idea of double-blind lineups (Wells, 1990) in, which are lineups conducted by someone who does not know which person in the lineup is the possible suspect and which are merely fillers. The idea of the double-blind lineup is to prevent inadvertent influence on the eyewitness from the lineup administrator. Double-blind lineup procedures are now required in many states and individual jurisdictions across the United States. In 2014, a report from a study committee the National Academy of Sciences endorsed the idea that all lineups should be conducted using double-blind procedures.

Wells also introduced a measure of lineup fairness called functional lineup size (Wells et al., 1979) to distinguish between the number of people in a lineup (nominal lineup size) and the ability of the fillers (innocent stand-ins) to make the lineup fair to the suspect.

In the 1990s, Wells and his Ph.D. student Amy Bradfield (Douglass) discovered that the confidence that eyewitnesses express in their identifications is highly malleable and can be dramatically inflated after making a mistaken identification through simple comments that seem to confirm their choice, a phenomenon known as the post-identification feedback effect.

In an influential article in 1978, Wells proposed a distinction between two different types of variables influencing the accuracy of eyewitness identification (Wells, 1978). System-variables are variables that are (or could be) under the control of the justice system (e.g., pre-lineup instructions to witnesses). Estimator-variables are variables that are not under the control of the justice system, but are circumstantial factors that influence identification (e.g. age, race).

Wells proposed that system-variable research would prove to have greater applied utility for criminal justice than estimator-variable research because system-variable research could inform the legal system of ways to reduce the chances of mistaken identification. Wells argued that the rate of misidentifications are influenced by several methodological biases in the methods used by law enforcement to secure the identifications The system-variable versus estimator-variable distinction that Wells introduced in 1978 has so thoroughly permeated the nomenclature of the eyewitness literature that the terms are now commonly used without attribution to their source.

Wells introduced the idea of using Bayesian statistics to describe eyewitness performance in eyewitness identification experiments (Wells, 2015). Wells has further developed these Bayesian methods to show how the amount of information (about guilt) gained from eyewitness identification evidence can be quite small and is highly dependent on other (non-witness) evidence.

Wells, along with Steven D. Penrod, also wrote a chapter on "Eyewitness identification research: Strengths and weaknesses of alternative methods" for Research Methods in Forensic Psychology. In this chapter, Wells and Penrod evaluate alternative research methods for eyewitness testimony, including laboratory and field experiments (Wells & Penrod, 2011). Wells and his colleagues collected field experiment data from four police departments and published the data showing that roughly 1/3rd of all identifications made by eyewitnesses resulted in the selection of an innocent stand-in lineup member (Wells, Steblay, & Dysart, 2015).

Wells worked with many states, starting with New Jersey in 2002 and North Carolina in 2003, to implement state-wide reform of their eyewitness identification procedures. Other states later followed in making reforms to eyewitness identification procedures based on the early models that Wells developed. Most states now require double-blind administration of lineups, pre-lineup instructions to witnesses warning that the culprit might not be in the lineup, the use of lineup fillers so that the suspect does not stand out, and the collection of a confidence statement from the eyewitness at the time of any identification. These same reforms have now been adopted by the International Association of Chiefs of Police.

Wells was the first to call into question the U.S. Supreme Court's ruling in Neil versus Biggers (1973) and its later reaffirmation in Manson versus Braithwaite (1977) about how to assess the reliability of eyewitness identification when there is suggestiveness involved in the identification. Wells' later writings have made an even stronger case for how the Court's approach is flawed given the findings in eyewitness science (Wells & Quinlivan, 2009). Recent state supreme court rulings have used Wells' critique of Manson vs. Braithwaite to fashion new approaches to assessing the reliability of eyewitness identification evidence (e.g., State v. Henderson in New Jersey and State v. Lawson in Oregon). In agreement with the basic arguments in the Wells & Quinlivan article, the report from the study committee of the National Academy of Sciences endorsed the idea of casting out the Manson v. Braithwaite approach to evaluating the reliability of eyewitness identifications.

Wells' testimony in court cases and eyewitness research on system and estimator variables has influenced legislation and state Supreme Court decisions. States such as New Jersey, North Carolina, Ohio, Vermont, Illinois, and Connecticut, for example, now require double-blind lineups and other safeguards for eyewitness identification evidence that Wells advocated. Wells' work with the U.S. Department of Justice under Attorney General Janet Reno resulted in the first set of national recommendations for law enforcement regarding the collection and preservation of eyewitness evidence (Eyewitness Evidence: A Guide for Law Enforcement).

In 2003, the United States Court of Appeals 7th Circuit upheld Wells' testimony in a Chicago civil suit pertaining to lineup procedures in which the defendant was pardoned innocence after the allegation that the police officer induced the three witnesses to identify him as the perpetrator (Newsome vs. McCabe, 2003). For his testimony, Wells conducted an experiment to examine the likelihood that all three of the witnesses would pick the defendant out of a lineup. The results suggested that the police officers manipulated the lineup due to the probability of picking the defendant being less than one in 1000. Reforms for legal standards in eyewitness identifications have been also noted in other cases such as State v. Larry R. Henderson and questions of constitutionality for eyewitness identifications have been raised (Perry v. New Hampshire).